On 2/8/24 11:20, Simon Josefsson wrote:
I would prefer having a global static variant since several uses of ctime is in non-thread global application context where even the above is unnecessarily complicated and wasteful of stack space.
I suppose we could add a static variant too, though this wouldn't be a stack space issue (the stack space is minimal).
A key point, is that these replacement functions won't follow ctime rules all the time. You can't reliably get a year number by looking at four particular bytes in the output. Any code doing that now will need to be rewritten no matter what.
